AI safety researcher

Srdjan Capkun

ETH Zurich

Google Scholar dblp ORCID Homepage

Last 12 months3 papers

AI safety papers from the last year

On this site: 3 papers

October 2026

Can Open Be Too Open? On the Security and Trustworthiness of Fully Transparent LLMs
Oct 26arXiv ↗

July 2026

MOSAIC: Masked Outsourcing of Secure AI Computations
Jul 31arXiv ↗

November 2025

Can LLMs Make (Personalized) Access Control Decisions?
Nov 25arXiv ↗

Authorship follows dblp and the author's own Google Scholar profile; only AI safety papers on this site are listed.